Optimizing Reactive Power Outputs of SVC After Faults in Renewable Power Grids

To address the issue of insufficient or excessive reactive power output in power systems equipped with a Static Var Compensator (SVC) during voltage sag and recovery events, this paper investigates the influence of SVC control parameters on reactive power output and thoroughly examines the corresponding voltage reactive power response characteristics. It is observed that, for the same set of control parameters K_s and dU, the objectives of maximizing SVC reactive power output during voltage sag and minimizing it during voltage recovery are inherently conflicting, thereby forming a multi-objective optimization problem. To solve this, a neural network is first trained using actual simulation data to construct a surrogate model capable of accurately simulating the transient reactive power output characteristics of most SVCs, overcoming the high computational cost and extensive iterative solving typically required by the Non-dominated Sorting Genetic Algorithm II (NSGA-II). The neural network is then integrated with NSGA-II for multi-objective optimization, resulting in the Pareto optimal front for the SVC's reactive power output, which significantly enhances SVC performance.